Documente Academic
Documente Profesional
Documente Cultură
2. Theory
The understanding of written text depends on three distinct components:
• Legibility
• Readability
• Comprehensibility
The first of these components is of no concern to us, as it is a responsibility of the layouters and
typesetters putting our writing into its final form.
The second of these we will deal with, as it is vital to having the reader actually read our document,
hopefully in full.
And lastly, the third component is essential to ensure that our reader will understand the purpose of
our writing.
These two components will be discussed in separate sections, even though some of the issues raised
may be pertinent to both.
In addition, we will also look at issues of style — some of writing’s do’s and don’ts , as even the
prose of technical writing does not have to be equivalent to a blunt axe when it might be an
instrument of precision.
Should you, kind reader, have suggestions for improvement to these pages, please let me know:
bernhard@icsharpcode.net
If a technical text is unreadable in the reader’s eye, he will quite probably assume that the product
described in this text also is of inferior quality. Code is a language, just as the language of the
documentation is. Not writing well in documentation implies faults in coding style.
In this case, we see that there is, as such, a larger than necessary number of commas.
Punctuation should be kept to a minimum. It is not necessary to put a comma wherever it looks
right. They often are not. Especially clauses of the ‘so that’ type can do perfectly well without
commas. This rule of course also holds for all other punctuation. And never, ever, try to transport
punctuation rules from your native tongue to the English you are writing.
It is given as a rule, which however is not the only such rule you may encounter, that sentences
should not exceed a desirable length of ten to fifteen words, never should fall below seven words or
extend beyond the ultimate limit of tolerable length reached at twenty words, even though longer
sentences may be found in high literature, where even punctuation as it is used in this example to
facilitate reading is oft omitted in novel experimental ways.
This of course is an example that runs somewhat longer than what you would expect to find in your
own writing. But read your own texts again and you will quite possibly find one or two of these
abominations, describing say, a complex chain of events and their handling. A complex train of
thought can only benefit from being broken down into sentences of convenient length. Temptation
to ramble on in one long sentence may be great. Resist. Your logic will benefit. Also cut out
anything not necessary to the immediate cause at hand. To quote Strunck and White’s third rule:
Sentences may be short. Then they are easy to read. And understand, too. But they look cheap. And
breathless. As well as leaving the reader restless.
Not much needs to be said here, as these above sentences illustrate the point to be made. If a thing
is worth saying, it is worth saying it well, not chopping language to pieces. Human language is not
a RISC language.
In general, try to vary the length of sentences in the limits given in the negative of subsection 3.2
above. Interesting writing depends on well dosed variations of length and choice of words — for
examples of the latter, see below.
3.4 Recursion
Sentences often turn into a quasi-circular case of recursion while reading them when the references
made in the sentence to the respective objects and subjects are left unclear by using the same
pronoun to describe these subjects and objects.
It is not easy to understand it when it is unclear what is referenced by ‘it’ – it by now should be
clear what it is supposed to mean, isn’t it?
Another form of recursion will take place in the reader’s mind when he is confronted with
convoluted sentences containing insertions, ellipses and other rhetoric figures. All of these will
need some place on his ‘stack’ — and a reader’s stack is shallow. ‘Pushing’ and ‘popping’ more
than three stack levels usually ends in disaster … for the message of your sentence!
The reader, that is, the intended recipient of our text, a hopefully clearly written and logically
structured document, will, if he is able to fully understand our prose, without difficulty come to a
safe assumption of what any given sentence, such as this a one, conveys, to him, the reader, by the
way of meaning.
The above sentence is grammatically correct, but will most probably provoke a ‘stack overflow’ in
our average reader. Such convoluted1 writing should be avoided wherever possible. Just as
recursive code, text may be ‘flattened out’. Take a monster such as the one above apart. Shorter
sentences make easier reading. If this is not possible, try to keep related parts of the sentences as
close to each other as possible.
Using the same words all over to describe the same things again and again is not pleasant even
more so when we can use different words to replace those same words we are using again and
again to describe the same thing in the same words.
For any given word, at least one synonym will be available. Do not hesitate to use a thesaurus.
Also, do not use the exact same phrasing again and again and again, unless it is intended to convey
some artistic intention — this however is almost never the case in technical writing. And:
Never use the same opening words in two or more subsequent sentences. Repetitive writing is
the enemy of all reader's interest.
4. Comprehensibility
In the complex process of reading, the step following the ‘tokenisation’ of the text is the actual
‘parsing’ — understanding what these symbols and their relations mean. A clear separation of these
two steps however can not be made.
A great portion of comprehensibility issues already was covered when we discussed recursion
1
Go ahead, look up ‘convoluted’ in a dictionary. Now. Do this as well for any other word you may not know in this
or any other text.
An understandable technical document always follows a logical structure. Any topic discussed is
based on the preceding topics. If a new concept is needed for the topic at hand, it needs to be
introduced before using it in dealing with this new topic. This holds true for any level of detail of
the document at hand, down to individual sentences.
The basic steps are:
1. Definition
2. Assumption/Theorem
3. Explanation/Proof
4. Conclusion
Of course, the classic structure of ‘thesis, antithesis, synthesis’ may be more appropriate for certain
topics, such as discussion of architectural decisions, but generally the above sequence is exactly
what we need.
On the ‘atomic’ level of a sentence, its logical structure is governed by raw grammar. Therefore, a
good working knowledge of grammar is absolutely necessary for getting our ideas across to the
reader as we mean them to be.
4.1 Definition
In this segment of the document, all terms and concepts necessary for the following should be
defined. In some cases, reference to previous material in the document is sufficient. Referring to
later sections is to be avoided at all cost. If it should for some reason be necessary, the definitions
may be placed in endnotes or a glossary at the end of the document. This should be clearly
indicated at the very beginning of the document. Footnotes are not intended for the purpose of
definitions. They are a place for further explanations or material of a ‘non sequitur’ nature2.
4.2 Assumption/Theorem
The task of this segment is the presentation of the idea or concept this particular document or part
of a document is supposed to deal with.
Make a simple and clear cut statement of where your argument starts and what will be the intended
outcome. No why or how should be given here. The why should be clear from earlier portions of
the document, the how is the subject matter of the next section.
In a user’s guide, this is where an indication of the function to be explained should be given.
4.3 Explanation/Proof
This segment of our document deals with giving justification for the idea put forth in the previous
section. It may be of purely argumentative3 nature — say, defending architectural decisions — or
come close to mathematical proof in style. In the case of a program implemented practically, this is
the place for explaining the workings of it step by step. Or in the case of a user’s guide, to explain
the interface and sequence of steps necessary for completing a given task.
2
Latin: ‘is not followed’ i.e. something that is not important for the understanding of the subsequent text, but merely
a side track of additional information.
3
Just read some classic rhetoric texts to understand what I mean. Socrates should ring a bell even without reading…
5. Matters of Style
This section will be concerned with the little things that will hopefully turn acceptable technical
writing into good writing. There is a number of seemingly small and unobtrusive ‘do’s’ and ‘don’ts’
that need to be watched out for consciously even though they seem to be obvious.
The following material is not sorted by relevance. You will just find things as they came to mind,
read through it and consider it as a wish list for good writing.
5.1 Title
The first thing our reader sees is the title. Therefore, the appearance of the title is of great
importance for the impression our text leaves. The dominating factor is capitalisation of the words
in the title. There are several 'schools of tought' as far as this is concerned:
Of these, I personally recommend number 44, as it is the style usually choosen for scientific
publications. Numbers 1 and 3 are usually bad for the readability of the title. In any, case, once you
have chosen a style, stick with it. No exceptions to the rule!
When numbering titles, consistency is also a must. If you settled for a certain scheme, stick with it!
Usually the numbering styles suggested by your text processing software – be that LaTeX, Open
Office or what ever else – of choice represent sensible schemes.
Don’t go above your station. It may well be tempting to employ vocabulary gleaned from some
obscure manual of language, obfuscating your intent by billowing clouds of rhetoric smoke, this
however will most certainly induct the reader’s total non-comprehension of your text.
By now it should be clear what the name of the game is: in the case of two given words of the same
meaning, preferably use the simpler one. With one exception: should the ‘bigger’ word be clearer as
to its meaning, use that.
And when you find yourself in the ‘Big Word Game’, also check the length of your sentences.
There seems to be a correlation between the two.
However, you will find that using only the smallest possible word to refer to a concept will making
your writing look dull. A bit of variety never hurts.
4
And use it for this Tech Note.
5.4 An 'a'
It is of course somewhat tempting when trying to write good English to use ‘a’ and ‘an’ in
accordance with the simple assumption that ‘a’ is to be used before consonants and ‘an’ before
vowels. This is not in accordance with the rules of proper English. Simple (assumed) rules usually
have exceptions to prove them in general. The exception in this case is with vowels:
• An apple
• An exception
• An idea
• A useless rule but
• An uppercase letter
This is of course a mere rule of thumb5. For the full story refer to a style manual such as ‘The
King’s English’.
5.7 Nativisms
In general, be suspicious of ‘too obviously natural’ translations of words from your native tongue
into English. E.g., the German ‘Konkurrenz’ will translate to English as ‘competition’, not as
‘concurrence’ as might be naively assumed. Get a good dictionary and check it often, read English
texts you are already familiar with in your tongue — no matter whether that text is originally
English translated to your native tongue or vice versa7. Preferably start with some favourite book of
yours — be that ‘The Lord of the Rings’, ‘Grimm’s Fairy Tales’ or ‘Gödel, Escher, Bach’ or even
comic strips or whatever else has taken your fancy.
Understanding words properly matters. When in doubt about the actual use of a word with several
potential meanings in translation, check with a native speaker or refer to a good reference
dictionary such as the ‘Oxford’. Some online resources for dictionaries are given in section 7.
7
The former case is preferable, of course.
8
As for flaming, IMHO is a pleonasm, IMO suffices perfectly well — no personal opinion is humble. This now
was a perfect example of a ‘non sequitur’ footnote.
5.12 Consistency
Be consistent in the use of either British or US-english spelling. Never switch inside any given
document. Examples of difference in spelling between those two are:
British US
Colour Color
Co-operation Cooperation
Customisation Customization
Also, once you choose a given technical term to mean one thing, use it only in that one sense. Do
not redefine!
6 Recommended Reading
Of course no complete list of books on style can be given here. The selection must by necessity be a
personal one. So I will just list a few books I found useful for myself:
Some books that are well written and will give some idea of what can be done:
Robert M. Pirsig, ‘Zen and the Art of Motorcycle Maintenance’ (An interesting book about
motorcycling, ,the human psyche, ‘quality’ and — a little — about technical writing, all
well written)
Douglas R. Hofstadter: ‘Gödel, Escher Bach – an Eternal Golden Braid’ (Some of the finest writing
ever on the concept of ‘Strange Loops’ which manages to get some rather technical points of
AI and programming across without being boring)
Donald Knuth: ‘The T X book’ (One of the finest software manuals ever written. Even the more
e
arcane aspects of TeX are clearly explained in an entertaining way)
The VAX manuals, a.k.a. ‘The Orange Wall’ (Tons of paper, the best set of software documentation
ever. Much better than most man pages and galaxies ahead of PC-software documentation)
7 Online Resources
Some online translation and dictionary resources:
http://www.tecnologix.net/laixicon/
http://dict.leo.org/
http://www-tgw.dfki.de/~winter/lang/anglizisms-de.html
http://www.linguadict.de/
http://www.dictionary.com/
http://wombat.doc.ic.ac.uk/foldoc/
http://www.onelook.com/browse.shtml
http://www.tuxedo.org/~esr/jargon/